IELTS Academic Reading Practice Test 138

In 1990 there was almost nothing to search. The World Wide Web, proposed at the CERN laboratory in Switzerland the previous year, was still being assembled, and most of the documents that universities and companies wished to share sat instead on FTP servers, machines that stored files for anyone to download. Finding a particular file among thousands of such servers was slow and uncertain, and it was this problem, rather than the web, that produced the first search tool.

Archie was written in 1990 by Alan Emtage, a graduate student at McGill University in Montreal, with help from Peter Deutsch and Bill Heelan. The name came from 'archive' with the letter v removed. Every so often the program contacted FTP sites around the world and downloaded their lists of files, combining them into a single catalogue that users could search remotely. Archie did not look inside documents at all: it collected the names of files, so a search could confirm that a file existed without saying whether it was worth reading. Even so, the service spread quickly, and imitations appeared within a year or two.

The next wave of tools was built for Gopher, a system developed at the University of Minnesota and released in 1991. Gopher presented the documents held on a server as a series of menus, through which users moved by choosing numbered items; it was simpler and more orderly than FTP. Two programs were written to search these menus: Veronica, created at the University of Nevada, and Jughead, written at the University of Utah. Neither could search the contents of the documents themselves, and Gopher's hierarchical structure was never designed for the chaotic linking that the web would allow.

By 1993 the web was growing fast enough to need indexes of its own. That year Matthew Gray, working at the Massachusetts Institute of Technology, wrote a program called the World Wide Web Wanderer, which travelled from one page to another and recorded what it found. Martijn Koster also released a tool called ALIWEB, which allowed the owner of a site to submit a short description of it. ALIWEB therefore depended on the honesty and energy of site owners, and it never attracted a large audience.

The decisive step was taken in 1994 at the University of Washington, where Brian Pinkerton wrote a program called WebCrawler. Like the Wanderer, it followed links automatically, but it also stored the words of every page it visited, and it was the first engine to index the full text of pages rather than merely their titles. In the same year a team at Carnegie Mellon University led by Michael Mauldin launched Lycos, which ranked its results by relevance instead of listing them in an arbitrary order; its name came from Lycosidae, the family of wolf spiders that hunt by running down their prey.

Also in 1994, two Stanford graduate students, Jerry Yang and David Filo, began compiling a list of interesting websites, arranged in categories. This was a directory rather than a search engine: every entry was chosen and described by hand, not by a machine, and as the number of sites grew the task became impossible to keep up with. Its popularity, however, was enormous, and for several years the most common way to begin looking for something on the web was to visit Yahoo and click through its categories.

The architecture that emerged during these years has changed far less than the machinery. A program known as a crawler starts with a modest list of addresses and downloads each page it reaches, following every link on that page to discover others, so the collection expands without human intervention. Software then extracts the words from each page and builds an index: a vast ordered record of which words appear where. A query never searches the web directly; it is checked against this index, and a reply can be produced in a fraction of a second. Finally, a ranking rule decides the order in which the matching pages are listed.

AltaVista, launched in December 1995 by the Digital Equipment Corporation, showed how well this design could work at scale. The company was one of the largest makers of minicomputers, and its engineers built systems that could hold an index of many millions of pages and reply almost instantly. The rules used for ranking, though, were crude: most engines counted how often a query's words occurred on a page, so a site that repeated a term hundreds of times, in text invisible to the reader but plain to a crawler, could push itself to the top.

The remedy came from Stanford, where two doctoral students, Larry Page and Sergey Brin, began in 1996 to build a system they called BackRub. Their insight was to treat a link from one page to another as a vote: a page was important if many pages pointed to it, and especially if those pages were themselves important. The method, PageRank, was much harder to fake than counting words, because a site could not simply insert extra links to itself. The two founded Google in September 1998, and within a few years the combination of a large index and link-based ranking had made their engine the most used in the world.

Almost every element of those early systems survives in modern search, though the scale would have been unimaginable in 1995.

Fresh water is abundant on Earth only in theory. More than 97 per cent of the planet's water is salty, and much of the remainder is locked in ice, so as cities expand and underground reserves are depleted, engineers have turned to a technology that separates water from the substances dissolved in it. At the centre of that technology is a membrane: a thin barrier, often no thicker than a human hair, that allows some substances to pass while holding back others. Reverse osmosis, the most important of these processes, now supplies drinking water to millions of people.

The principle is simple enough. Salt water is pressed against one side of a membrane that water molecules can cross but dissolved salts cannot. Left alone, water would move the other way, from the dilute side to the concentrated one, until the pressure on the salty side was great enough to stop it; that pressure is called the osmotic pressure of the solution. Applying a pressure greater than the osmotic pressure reverses the flow, drawing water through the membrane and leaving the salts behind. The name of the process follows from this reversal.

Membranes are classified by the size of the particles they hold back. Microfiltration and ultrafiltration, with pores measured in tens or hundreds of nanometres, remove bacteria, suspended solids and large organic molecules, and are widely used to prepare water before it reaches a finer membrane. Nanofiltration, with pores of about a nanometre, removes most multivalent ions and many organic compounds. Reverse osmosis membranes have no pores in any ordinary sense: water dissolves into the dense polymer and diffuses through it, while ions are excluded both by their size and by the electric charge of the material.

The modern era began in 1960, when Sidney Loeb and Srinivasa Sourirajan at the University of California, Los Angeles, produced a membrane of cellulose acetate that passed far more water than earlier designs while still rejecting salt. Their trick was to make the membrane asymmetric: an extremely thin, dense skin supported by a much thicker, porous layer that offered little resistance to flow. Almost two decades later, John Cadotte, working for the company FilmTec, developed a composite membrane in which a polyamide skin a few tens of nanometres thick was formed directly on a porous support by interfacial polymerisation, a reaction between two liquids at their boundary. This remains the basis of most reverse osmosis membranes in use today.

The cost of pushing water through a membrane is dominated by energy. Because seawater has an osmotic pressure of roughly 27 bar, a plant must apply a pressure well above that — typically between 55 and 70 bar — and the pumps that do so consume a great deal of electricity. In the 1970s, desalination required more than ten kilowatt-hours for every cubic metre of fresh water produced; today's most efficient plants use around three or four, helped by devices that recover pressure from the stream of concentrated brine leaving the system. A plant on Israel's Mediterranean coast that began operating in 2013 produces more than 600,000 cubic metres a day.

Membranes have weaknesses. The most serious is fouling: particles, organic matter and the growth of bacteria form a layer on the surface, which reduces the flow and eventually forces the plant to stop and clean the membrane. Polyamide is also attacked by chlorine, the cheapest and most convenient disinfectant, so the water must be carefully dechlorinated before it reaches the membrane and disinfected afterwards — an awkward and costly arrangement. The brine that remains, often about half the volume of the feed water and roughly twice as salty as the sea, must be disposed of, usually through a pipe that spreads it across the seabed, where its density and warmth can affect marine life.

Desalination gets the attention, but most membranes are used for something else. Haemodialysis, which cleans the blood of patients whose kidneys have failed, depends on membranes that let small waste molecules through while retaining proteins and blood cells. The dairy industry uses ultrafiltration to concentrate milk protein, and breweries and soft-drink manufacturers rely on reverse osmosis to remove minerals that would affect taste. Treating municipal wastewater to a standard high enough for irrigation, or even for drinking, is now routine in Singapore and several other water-scarce places.

Research is now directed at the three weaknesses of energy, fouling and durability. Membranes incorporating graphene oxide, whose sheets can be stacked to create precisely controlled channels, promise higher flow at the same pressure, while biological membranes, which achieve remarkable selectivity using aquaporin proteins that pass water and almost nothing else, have been imitated in the laboratory. Forward osmosis, in which water is drawn across a membrane by a concentrated solution rather than pushed by pumps, may suit places where electricity is scarce, and membrane distillation, which uses waste heat, is being tested in the Gulf states. Whether any of these will displace polyamide on a large scale is still unclear.

What is certain is that the membrane, a device with no moving parts, has quietly become one of the most important machines of the modern world. A technology that produced a few litres a day in a laboratory in 1960 now treats tens of millions of cubic metres daily, and the limits of the process are set less by chemistry than by the price of energy and the discipline of maintenance.

Cryptocurrencies were designed to make regulation difficult. Their founding document, a short paper published in 2008 under the name Satoshi Nakamoto, described a system for transferring value between strangers without a bank, a central authority or any institution capable of blocking a payment. Fifteen years later the same assets are traded by some of the world's largest financial firms. The question is no longer whether governments should take an interest, but what form it should take; and on this the industry's libertarian origins offer little guidance.

The technology is easier to describe than to regulate. A blockchain is a ledger maintained simultaneously by thousands of computers, each holding a copy and updating it according to agreed rules, so that altering the record would require an implausible amount of computing power. The ledger records which addresses hold which assets; it says nothing about who controls those addresses, since they are pseudonymous. This combination — a public record of transactions and a private record of identities — makes the technology attractive both to people who value financial privacy and to those who wish to evade the law.

The arguments in favour deserve to be taken seriously. In countries where inflation has destroyed savings, or where sending a small sum home costs a tenth of a modest wage, an asset that can cross a border in minutes for a few cents answers a genuine demand. For people whose accounts can be frozen by a hostile government, holding value outside the banking system is a form of protection rather than a plaything. Programmable payments, which move only when stated conditions are met, may eventually cut the cost of trade finance, though the examples so far are modest.

The arguments against are equally substantial, and they involve sums that dwarf the benefits just described. The prices of the largest cryptocurrencies have repeatedly halved within months, and because the assets are marketed to ordinary savers as investments, those falls have been paid for by people who could least afford them. Ransomware attacks depend on payments that are difficult to reverse, sanctions are weakened when designated entities hold value in wallets no bank controls, and mining, which secures some networks, consumes electricity on the scale of a medium-sized country.

Yet the most instructive failures have not been failures of the technology. When the Tokyo exchange Mt Gox collapsed in 2014, when the Terra project disintegrated in 2022 and when FTX, then one of the largest exchanges, filed for bankruptcy later the same year, the cause in each case was not a flaw in the code but the conduct of people who had gathered customers' money and lost it. FTX's founder was subsequently convicted of fraud. These were conventional failures of custody, disclosure and governance, in a market that had placed itself beyond the reach of the regulators who normally police such matters.

That observation cuts against both camps. For enthusiasts, it is a reminder that a system designed to remove trust from finance has, in practice, reintroduced it in the least accountable form imaginable: the customer must trust an exchange, a stablecoin issuer or a lending platform, without the protections that trust normally brings. For those who would simply ban the assets, it is a reminder that the harm was concentrated in businesses rather than in the ledger, and that businesses can be supervised.

The claim that regulation would destroy innovation is, in my view, the weakest of the arguments. Innovation in financial services has always required law: limited liability, securities markets and payment systems were all created by legislation, and firms in those industries operate profitably within dense rulebooks. What destroys innovation is uncertainty, in which a legitimate business cannot know whether its product will be permitted next year. Clear rules on the treatment of customer assets, on the information given to buyers and on the custody of private keys would allow serious firms to invest with confidence.

The alternative, prohibition, has been tried. China banned trading platforms in 2021 and drove the activity offshore; the trading did not stop, it merely moved to venues that reported nothing and offered customers no recourse. A country that outlaws an asset also surrenders any influence over the standards that govern it, while still exposing its citizens to the risks.

What, then, should rules do? They should concentrate on the points where cryptoassets meet the ordinary financial system. An issuer of a stablecoin — a token advertised as holding a fixed value — should be required to demonstrate that it holds adequate reserves, since a stablecoin that is not stable is a promise that has quietly been broken. Exchanges that hold customers' funds should be subject to rules on segregating those funds and on how they are kept. Lenders should disclose their exposure, so that a collapse in one part of the market does not surprise another. Regulators will need the 'travel rule' to follow funds between wallets, and the energy consumed by mining deserves attention too.

None of this requires regulating the software, which would in any case be futile; a blockchain protocol cannot be summoned to appear before a committee. It requires regulating the institutions that stand between the public and the technology: the exchanges, the issuers, the brokers and the banks that lend against the assets.

My own view is that cryptoassets will not replace money as their most fervent advocates once promised, and that they will not disappear either. They will occupy a small, useful and occasionally dangerous corner of finance. The task of regulation is not to choose a side in that argument but to ensure that the public bears only the risks it has chosen to bear.